Since we started AptoNow in April last year, I’ve spent the last 12 months engaging intimately with higher education timetablers from across Australia, New Zealand and the UK. Countless hours have been spent deeply immersed in intricate and extensive timetabling datasets. My biggest insight over this time? There are huge gains in student experience and resource efficiency that can be unlocked by using data in three key ways.
First, leverage data on historical student behaviour to unconstrain the timetable.
University timetables are often over-constrained. Working with a top-ranked partner IRU university, we found that approximately half of clash constraints in the timetable had no basis in student behaviour (see chart below). This places huge restrictions on what timetablers can do to meet stakeholder demands.
An over-constrained timetable is not only more challenging and time-consuming to build—ironically, for the extra effort, staff and students get a worse timetable. Why is this? Over-constraining a timetable means that any needed downstream alterations will have significant cascading effects. The timetable thus becomes inflexible to change and highly fragile when changes are necessary.
The upfront work to build an appropriate and negotiable analytical framework around timetable constraints can save hours and days of work later in the build, as well as make room for other quality initiatives.
Student choice is critical for modern universities, but it is not the only factor driving satisfaction and experience of the timetable. How are lectures and tutorials sequenced across a week? How far must students travel between classes, and how long must they wait on campus between their morning and afternoon schedules? All these considerations get crowded out when a timetable has been over-constrained in the first step of the build.
Second, use class registration data to understand student timetable ‘clustering’.
Timetable clustering refers to how condensed (or dispersed) individual students’ schedules are throughout the week or within a day. For example, does a student have on average two classes each day in the morning, or alternatively two full days of classes on Monday and Tuesday? Working with a top Go8 university, we found that the average student has a fragmented timetable, with classes across 3.5-4 days per week. However certain programs managed to be much more clustered, fitting the same amount of instruction into just 2 days per week (see chart below)!
This is a great proof-of-concept for what is possible across student cohorts. As it stands, most institutions do not have explicit objectives regarding timetable clustering, so they often lack robust diagnostic tools to understand how they are doing on this front.
Student cohorts may differ in preference across university campuses, across disciplines, or across particular programs within disciplines. If your institution wants to respond to student feedback or complaints about their individual schedules, then you need benchmarks and clear anchor points for discussion.
In my experience, universities want to address valid concerns in this area, but it’s just difficult to manage. They need systematic ways to distil the complexity and reduce the problem into bite-sized chunks. This can allow for (A) a reliable diagnosis of the issue and (B) a gauge to set reasonable targets and understand real effects of initiatives.
Third, leverage automation to approach the timetable in a more strategic way
Timetablers have told us repeatedly that the Holy Grail of university timetabling is a single button that, once pushed, pops out an entire timetable that meets required quality standards. Because of how previous autoschedule solutions have performed, however, there’s a great deal of scepticism around the quality of a fully autoscheduled timetable. What sorts of hidden errors will be lurking, and how many issues will it ignore and leave for staff to deal with manually anyway?
Technology and data science, however, have come a long way in the last decade. Once we started implementing autoschedule solutions with our partners, we immediately recognised its true value: Strategic planning and student success.
Under the current paradigm, many stakeholders see autoschedule simply as a potential time saver—cutting down on their manual effort. But the lion’s share of the benefit is not in the time and effort saved. It’s in the added value and ability to reliably simulate effects, feasibility and implications of major policies and initiatives. Once an autoschedule algorithm can be fined tuned to sufficiently meet quality standards, institutions can begin to quickly experiment with the nuts and bolts of how they deliver education.
For example, what if the university needs to take an entire building off line for renovation? Where are the critical bottle necks going to be? How feasibly can we change our academic calendar from semesters to trimesters? What are the key trade-offs between pursuing a space optimisation strategy vs. a timetable clustering initiative? Once institutional parameters and requirements are loaded, these are the exact types of scenarios that a reliable autoschedule solution can simulate.
Reach out to us below.
At AptoNow our central objective is to change how timetabling works. We want to empower timetabling teams with the tools to, not only efficiently build academic schedules, but also to provide university leaders with the information they need to deliver on key strategies and objectives that benefit the range of institutional stakeholders.